While enormous progress has been made in elucidating genetic factors underlying autism spectrum disorder, it is largely unknown how genetic and non-genetic risk factors integrate and how they shape severity of social communication and cognitive deficits. This gap can be addressed by developing comprehensive liability models using a population-based epidemiological sample with dense genetic and phenotypic data. To fill this gap, we have developed the Population-based Autism Genetics and Environment Study (PAGES), involving a Swedish epidemiological cohort obtained by ascertaining samples with DSM-IV autistic disorder (AD), which captures more severely affected individuals, chosen from a national, population-based sample of over 7,000 living individuals. Modeling liability in the epidemiological sample of AD has provided accurate estimates of the risk conveyed by common and rare genetic variation. This study has also revealed that ~40% is still unaccounted for. Combining critical environmental variables (paternal and maternal age, gestational history) and phenotyping data (IQ, autism severity, family psychiatric history) with measures of heritability is key to fully understand autism liability. We now propose to strengthen PAGES by pursuing the following specific aims: 1) To recruit, genotype and sequence at least 1,500 additional cases, including 1,350 less severely affected individuals; 2) To study common and rare genetic variation in relation to ASD severity and cognitive function; 3) To determine how other sources of putative risk for ASD are distributed in relation to ASD severity and cognitive function, and, 4) To discover risk genes for ASD by analysis of whole-exome sequence data and identify common risk variation by genome-wide association study (GWAS). We expect to contribute liability models that integrate genetic and environmental risk factors and take into account the phenotypic complexity along two core dimensions: severity of social deficits and cognitive function. In our opinion, this is significant because it allows us to: 1) study rare genetic variation at all scales across phenotypic groups; 2) understand the interplay between polygenic risk and highly penetrant rare variants across phenotypic groups; 3) measure heritability and environmental influences in light of phenotypic variability; and, 4) define the familial burden, both genetic and non-genetic. In our opinion, our study is innovative because it probes specific components of risk, both genetic and non-genetic, in a population-based cohort and introduces phenotypic variability as an additional dimension. It is also innovative because it combines genetic (additive and rare inherited) variation, parental age, and family history of psychiatric disorder to assess the familial burden, while introducing novel tools and approaches to genetic analyses. This radically new way of tackling ASD liability, compared with current studies, will provide novel insights into autism risk factors and their interactions in determining phenotype, thus opening new avenues for clinical assessment of risk, prevention and clinical care.